Opened 3 years ago

Last modified 2 years ago

#16797 new bug

DHCP_REQUEST loop of no link...reconfiguring... ready... no link ...

Reported by: AlienSoldier Owned by: axeld
Priority: normal Milestone: Unscheduled
Component: Servers/net_server Version: R1/Development
Keywords: Cc:
Blocked By: Blocking:
Platform: All

Description

For quite sometime i have this occuring at least once or 2 per month. Can't seem to see what trigger it.

Restarting my router don't change a thing neither is reasking a new DHCP communication from networking pref panel, only a computer reboot fix this.

Linked is a syslog of a typical occurance.

I got this on more than one computer but it seem to be driver related as on some i never see the problem.

Attachments (1)

DHCP_REQUEST (41.0 KB ) - added by AlienSoldier 3 years ago.

Download all attachments as: .zip

Change History (8)

by AlienSoldier, 3 years ago

Attachment: DHCP_REQUEST added

comment:1 by diver, 3 years ago

Component: Network & InternetServers/net_server
Owner: changed from nobody to axeld
Version: R1/beta2R1/Development

Please always mention haiku revision.

comment:2 by AlienSoldier, 3 years ago

well it is since a long time ago. Last time it happened it was on my previous one: hrev54940. I do not have my current one hrev54955 to know if it can happen on that one.

comment:3 by ttcoder, 3 years ago

DAEMON 'DHCP': /dev/net/rtl81xx/0: Timeout shift: 4294249842 msecs (try 6)

#15723 shows similar lines though they have 3 or 4-digit timeout shift values, whereas in this one the "4294249842" value is huge (it's 0xfff50d72 in hex)

comment:4 by tqh, 3 years ago

Cisco docs say that NACK's are most likely to happen if there is more than one DHCP server on the network. Easiest way to check that is probably with wireshark. Our DHCP client could also keep track of what IP is responding, and warn if there is a change in IP.

I need to look at code a bit to refresh mem, but that number looks weird.

comment:5 by maquak, 3 years ago

I have similar problem with this driver (rtl81xx) and it happens when I cold boot into Haiku. Restarting my PC doesn't help, but when I boot Windows, it connects without problem and then when I reboot to Haiku, Haiku is working well too, so I have a feeling that Windows driver sets some state that Haiku driver does not.

The difference for me is that I don't have this overflow values for DHCP timeouts.

My setup:

  • hrev54982 x86_64
  • Realtek® 8111H Gigabit LAN Controller - this is on MSI X570-A Pro motherboard

There are other tickets that might be caused by the same problem: #15866, #16731.

comment:6 by waddlesplash, 2 years ago

I note that Realtek has their own FreeBSD driver for this class of hardware, separate from FreeBSD's upstream version. I think when I looked at it once, the internals looked very different from FreeBSD's.

The one thing I see in the posted syslog that stands out:

KERN: [net/rtl81xx/0] watchdog timeout

comment:7 by AlienSoldier, 2 years ago

Happened one more time, this time on hrev55726.

Note: See TracTickets for help on using tickets.